Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Here's one example

Author: Robert Hyatt

Date: 15:03:52 08/06/02

Go up one level in this thread


On August 06, 2002 at 01:00:04, Robin Smith wrote:

>On August 05, 2002 at 17:50:27, Robert Hyatt wrote:
>
>>On August 05, 2002 at 12:15:32, J. Wesley Cleveland wrote:
>>
>>>On August 05, 2002 at 03:10:07, Jouni Uski wrote:
>>>
>>>>[D]8/4rk2/1p2r3/1Pp5/2Pp4/1K1P4/2PQ4/8 w - -
>>>>
>>>>Crafty's evalution after 14 ply +1,56.
>>>>
>>>>Jouni
>>>
>>>Crafty 18.15 static evaluation
>>>material evaluation.................   1.60
>>>development.........................   0.00
>>>pawn evaluation.....................  -0.04
>>>passed pawn evaluation..............   0.00
>>>passed pawn race evaluation.........   0.00
>>>king safety evaluation..............   0.00
>>>interactive piece evaluation........  -0.14
>>>total evaluation....................   1.42
>>>
>>>
>>>Crafty 18.10 static evaluation
>>>material evaluation.................   0.80
>>>development.........................   0.00
>>>pawn evaluation.....................  -0.04
>>>passed pawn evaluation..............   0.00
>>>passed pawn race evaluation.........   0.00
>>>king safety evaluation..............   0.00
>>>interactive piece evaluation........  -0.14
>>>total evaluation....................   0.62
>>>
>>>Apparently Bob does believe that Q+P is much better than 2R
>>
>>
>>I believe that in most positions, unless something really unusual is
>>happening, that a queen is better than two rooks, when the computer has
>>the queen.  As a general rule, the queen can _always_ force a draw,
>>because of the many checks it can give.  And it often finds ways to pick
>>up a pawn here and there.  The exceptions occur when the rooks get doubled
>>and can't be separated, but even then it is not a bad idea.
>>
>>I simply count a queen as equal to two rooks, period...  And in 99.9%
>>of the cases, that is at least correct...
>
>99.9%?  999 times out of 1000 the queen is stronger than two rooks??  I find
>this very hard to believe, even in computer games.  In general, deciding whether
>the queen is stronger or the two rooks are stronger is impossible without
>knowing more, in particular knowing about the pawns and if the rooks are
>protected.  If the pawns are loose and scattered and/or the rooks are loose, the
>queen can do well, since the rooks can't defend them all and the queen can pick
>them off.  But give the rooks some protection and connected and defendable pawns
>and the rooks can win much more often than the queen will.  Perhaps what you are
>seeing is an artifact of the kinds of pawn structures that arise in comp-comp
>games, and not a feature of the relative value of queen versus rooks in
>comp-comp games.
>
>By the way, the original position posted is so utterly drawn that seeing it as
>+1.42 is pretty funny.
>
>Robin


What I said, or what I _meant_ to say, is that in 99.9% of the games, the
queen is _no worse_ than the two rooks.  Which is why I consider having a
queen equal to having two rooks roughly.  Queen and pawn vs a pair of rooks
will have the queen side better...

I agree that in the posted game, a human says "all pawns on one side, this
is a draw".  But that is just one of many different possibilities.  Considering
the queen less than two rooks leads to even more mistakes.  I try to keep the
error term to a minimum, since zero is not possible.

Note that this is simply my opinion, based on watching a _bunch_ of ICC games
over the years.  With my program playing both sides.  As always, your results
may well be different...  and my opinion is certainly subject to change if I
see something to change my mind...



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.